G11 Tracking Error Characterization

N.B. PE and Period are eyeball estimates from the graphs. RMS values are calculated from the raw data after removing any drift with the appropriate polynomial fit.
Click on the date to view the drift corrected tracking data (if available), click on the drift log tag to uncorrected tracking data.

Date	PE (")	RA RMS (")	Period (sec)	DEC RMS (")	Config	Comment
3/28/2006	10	2.7	30	0.5	Original configuration	Displays five peaks 30 sec apart, with a 1 min section that is either flat or sinusoid depending on relative phase or two components. Five worm cycles have been overlaid with a Savitzky-Golay filter (21pts, 2nd order).
4/1/2006	10	2.5	76	0.55	Reemed out RA Oldham coupler	Log was acquired while/after ST7 imaged M13, therefore MX716 was the guide camera, slow update rate (2-3 sec) therefore I assume MX716 was in binned mode. Classic 4 or 5 30sec peaks with flat spot is swamped by 76sec oscillation.
4/2/2006	10	2.9	76	0.77	Tried to improved coupler alignment, or perhaps not!?	Improved alignment was no improvement, randomness cycle to cycle might even be worse.
2006-04-09 023546						Data is the same as the previous run, only more random. Noticed that coupler was slipping on shaft during slews so packed it in, I believe this explains some of the randomness.
2006-04-09 015834	16	4.1	76	0.5	Helical coupler	I vertically displaced the traces on the PEraw chart to reveal the dominant 76sec oscillation sliding (sometimes) with respect to the worm phase.
2006-04-09 223103 drift log	10	2.9	76	0.46	Replaced original unmodified Oldham coupler. Swapped original DEC worm into RA axis (leaving worm blocks alone)	Vertically displaced traces on the PEraw chart reveals the dominant 76sec oscillation sliding (consistently) with respect to the worm phase.
2006-04-10 003959 drift log	8	2.2	30/76	0.65	Swapped original DEC motor/gearbox into RA axis	Sliding 76 sec oscillations very low amplitude but still visible. Occassional 30 sec peaks now dominate with a flat spot of about 1 min.
2006-04-17 011210 drift log	9	2	25/30	0.64	Enlarged RA worm bearing holes and installed ABEC5 bearings	Sliding 76 sec oscillation non existant according to PEMpro, however 80, 120 sec harmonics now apparent. Sharp slopes spaced every 25-30sec over 2.5 min period are the worst offender.
2006-04-17 223601 drift log	9	2.1	35	0.69	Swapped worm gears (original RA worm into RA axis)	Sharp slopes spaced every 35sec over 3 min period are the worst offender.
2006-04-18 215838 drift log	10	2.1	35	0.69	Realigned RA Oldham and worm.	Sharp slopes spaced every 35sec over 3 min period seemed to be reduced.
2006-04-19 234447 drift log	10	2	35	0.9	RA axis cleaned/lubed	Results look identical to 2006-04-17 223601. How is that possible? Unless they are both the best possible alignment for this worm.
2006-04-21 005327 drift log	11	2.5	35	0.66	Best adjustement of Oldham so far, out 0.016" horizontal and < 0.002" vertical	Shape consistent with best alignment, however curve is noticeably noisier.
2006-04-21 013226 drift log	11	2.4	35	0.7	Intentionally misaligned RA Oldham coupler 0.025" vert., and 0.015" horz.	Shape and noise level about the same as previous attempt.
2006-04-25 225334 drift log	9	1.9	35	0.62	Milled worm block knob so that alignment is now <0.005" horizontal and <0.002 vertical.	Slight improvement, however basic contaminant remains.
2006-04-26 012123 drift log	8	2	35	0.93	Decoupled worm from Oldham coupler for 1min while motor running, then reengaged.	Pattern same as previous except shifted by 1min, this indicates that sharp slopes are solely from the worm.
2006-04-27 223347 drift log	5	1.5	240	0.5	New worm	Much improved result, autoguiding now works!

Analysis

dError/dt before (2006-04-25-225334, blue trace) and after (2006-04-27-223347, magenta trace) worm replacement. Vertical scale is the derivative of the tracking error in arcsec per second. I chose these two to compare since they had similar RMSde values.

Notice the peak dError/dt goes from 1 "/sec to less than 0.4 "/sec while the change in PE is less (9" to 5") and the change in RMSra is even smaller (1.9" to 1.5"). The difference in the quality of the astrophotos from before and after is substantial, from oval stars 100% of the time to round stars > 80% for exposures as long as 10min. I believe this demonstrates the how maximum slope error is a more crucial parameter to control as compared to PE (or RMSra for that matter).

Improvements to the G11 drive train.

Helical versus Oldham Coupler on the G11

Unguided Smear Model

Screenshot of a zipped Excel spreadsheet that models the probability of getting an unguided image with stars of a given ellipticity (or flatness = Ellipticity - 1). For a properly sampled seeing disk the flatness parameter should be less than ~0.2 to give pleasing stars. For very undersampled images (e.g. > 3 "/pix) the tolerable flatness parameter can be much higher.

High Frequency Noise in a G11